52406_07a

. . . . . . . . . . . . . . CHAPTER 7

Solution Purification and Concentration

The processes considered in this chapter selectively concentrate dilute (e.g., 1 to 20 g Au/t) gold-bearing solutions to produce a higher-grade solution from which gold can be extracted most efficiently by the recovery methods described in Chapter 8. Gold is first adsorbed from leach solutions onto an extractant, such as activated carbon or a synthetic ion exchange resin. The loaded extractant is then separated from the process stream, and the gold values are desorbed into a smaller volume of solution suitable for metal recovery. The stripped extractant is regenerated, if necessary, and then reused in the process. Activated carbon is the most widely used extractant for this purpose. Alterna- tively, ion exchange resins have been used in some applications and continue to be developed. Both extractants can be used to treat leach slurries directly, called in-pulp processing, as well as unclarified and clarified solutions, thereby obviating the need for solid-liquid separation steps required in conventional flowsheets.

Liquid solvents have also been investigated for use in gold extraction and are considered briefly.

7.1 CARBON ADSORPTION

7.1.1 Properties of Actlvated Carbon

Activated carbon, or charcoal as it is now less commonly called, is an organic material which has an essentially graphitic structure. Due to a highly developed internal pore structure, it has an extremely large specific surface area, and values in excess of 1,000 m2/g are not uncommon. As a result, activated carbon has found diverse industrial applications in both gas and liquid separation processes; however, its use in the gold recovery industry has only been widespread since about 1980 (Chapter 1).

The most important properties of activated carbon for use in gold extraction are the following: . Adsorptive capacity . Adsorption rate . Mechanical strength and wear resistance . Reactivation characteristics

Particle size distribution Other nontechnical considerations such as cost, availability, and service by supplier also affect carbon selection.

The properties listed here are considered in more detail in Section 7.1.2; however, it is first necessary to consider the factors that determine these properties, namely the type of source material and the methods of manufacture and activation.

297

298 I THE CHEMISTRY OF GOLD EXTRACTION

7.1.1.1 Manufacture and Activation If properly treated, virtually any carbonaceous material can be used to produce activated carbon. The most commonly used source materials are wood, peat, coconut shells, bitu- minous coal, anthracite, and fruit pips. The type of source material has a marked influence on the physical structure of the product, in particular, the pore volume and particle size distribution. For example, wood is used as the source material for decolorizing carbons, whereas coconut shells and coal-based carbons are generally used for gas phase adsorption and gold recovery applications. By the 1980s, some activated carbon was being manufactured by extruding powdered and agglomerated peat- or coal-based source materials to produce similar sized pellets.

Carbon is activated by removing hydrogen, or hydrogen-rich fractions, from a carbonaceous raw material to produce an open, porous residue. This process is called activation, and is typically achieved in two stages (Figure 7.1).

In the first stage, the material is heated to approximately 500°C in the presence of dehydrating agents, a process called carbonization. Many of the impurities are removed as gases (e.g., carbon monoxide, carbon dioxide, or acetic acid) or remain as a tar-like residue on the carbon. As a consequence, carbon atoms are freed to some extent and group together as crystallographic formations, known as elementary crystallites. This results in the development of a product with a specific surface area between 10 and 500 m2/g (and sometimes as high as 1,000 m2/g), the majority of which is due to micropore formation [2,31.

The second stage consists of exposing the carbonized material to an oxidizing atmosphere of steam, carbon dioxide, and/or oxygen (air) at temperatures of 700 to 1,OOO"C to burn off the tar-like residues and to develop the internal pore structure. Further reaction results in partial or complete burnout of carbon layers, producing a widening of existing pores and exposing the surfaces of the elementary crystallites formed during carbonization. Carbon atoms at the edges and corners of the elementary crystallites, and at defects or discontinuities, are especially reactive due to their unsaturated valencies and are called active sites.

The reaction of steam with carbon is thought to first involve the adsorption of water vapor onto the carbon surface, followed by the evolution of hydrogen and carbon monoxide. The reactions that follow are postulated, where a square bracket (1) indicates bonding with the carbon surface [4, 51.

]C + H,O + lC(H2O) + CO --L lC(0) + H, II co (EQ 7.1)

This mechanism is represented by the overall reaction:

C + H,O + H, + CO AGO = -130 kJ/mol (EQ 7.2)

This process is sometimes applied during carbon reactivation in gold extraction plants by the addition of steam into reactivation kilns (see Sections 7.1.4.4 and 7.1.5.7). The activation of carbon is catalyzed by iron, copper, and oxides and carbonates of the alkali metals.

Although coconut shell carbons are the most commonly used activated carbons in the gold extraction industry, extruded (peat-based) carbon has been used increasingly, and there is potential for other source materials to provide alternative carbons in the future, for example, peach and apricot pips and sugar cane residue [6, 71.

In the early 2000s, magnetic activated carbons (MACs) were produced by mixing a magnetic precursor material (e.g., iron citrate) with a suitable carbon source (such as

[CH. 7 SEC. 7.11 SOLUTION PURIFICATION AND CONCENTRATION I 299

Wood Charcoal Kneading

Grinding 1 Tar

/ Drying . -

Extrusion Ageing

Classification I 9-G Carbonization Gas I -

G a r i s t e a m - 9 n

E n n u \ Activation

Classification

Packing

FIGURE 7.1 Flowsheet for production of active carbon by activation with steam [l]

pinewood) and heat treating the mixture under controlled conditions of temperature and gas phase composition. Investigation and evaluation of activated carbon produced in this way have indicated the presence of micropores as well as mesopores during the early stages of activation [8 ] . MACs show potential for future application and optimiza- tion of carbon adsorption systems, with the potential for faster adsorption kinetics and the ability to separate carbon from solution or slurry by wet, high-intensity magnetic separation.

7.1.1.2 Physical Properties Activated carbon has a similar, though less well-ordered, structure to that of graphite. X-ray studies have suggested that activated carbon has two basic structures [41: . Small regions of elementary crystallites, composed of roughly parallel layers of

hexagonally ordered atoms . Disordered, cross-linked, spaced lattice of carbon hexagons, which is more pro- nounced in chars formed from materials of high oxygen content.

Commonly quoted dimensions for elementary crystallites are temperature dependent, but typically vary from 9 to 12 high and 20 to 23 A wide. From these values it has been estimated that the crystallite structures are approximately three layers high, with widths equivalent to the diameter of nine carbon hexagons.

The activation process generates an extremely large internal surface area, which is practically infinite relative to the outer surface of a carbon granule, and a wide range of pore sizes and shapes. Unfortunately, because it is not possible to accurately determine the shape of pores, this leads to some difficulty in expressing pore size. The classification of pore size by Dubinin [ 5 ] is generally accepted and is based on changes in gas or vapor adsorption mechanisms with pore size: . Macropores: x > 100 to 200 nm

Transitional pores or mesopores: 1.6 < x < 100 to 200 nm . Micropores: x < 1.6 nm where x is the characteristic size. The term “supermicropore” has been used to describe the range 0.6 to 1.6 nm [91.


The main distinction between gas-adsorbing and decolorizing carbons lies in their pore size distributions (Figure 7.2). Coal-based carbon has a large number of mesopores, which is important for adsorption kinetics as they allow access to micropores. This is significant because the gold cyanide anion is relatively large and may be inaccessible to as much as 90% of the carbon micropores. However, coal-based carbons are not as mechanically strong as coconut shell carbons and are less suitable for gold extraction systems.

Activation with carbon dioxide can produce carbon with smaller pore volumes and a greater proportion of micropores. On the other hand, activation in the presence of oxygen only develops porosity to a limited extent due to pore blockage by surface oxides [ 9 ] . In many other applications, including gold cyanide adsorption, transportation of the adsorbate within the pores can be rate determining. Consequently, decreasing the particle size of granular carbon has a large effect on the rate of adsorption, despite only a small increase in net surface area.

Typical coconut shell activated carbon has an ash content of 2% to 4%, an apparent density of 420 to 450 kg/m3, BET surface area of 900 to 1,000 m2/g, and an iodine number of 1,100 to 1,200 mg 12/g.

7.1.1.3 Chemical Properties The adsorptive properties of activated carbon are not only determined by surface area but also by its chemical properties. Although these characteristics are less well under- stood, the activity of carbon is attributed to the effects listed: . Disturbances in the microcrystalline structure, such as edge and dislocation

effects, which result in the presence of residual carbon valencies. This affects the adsorption of both polar and polarizable species. . The presence of chemically bonded elements such as oxygen and hydrogen in the source material or chemical bonding between the carbon and species in the acti- vating gas. The nature of chemically bound oxygen and hydrogen is dependent upon the type of source material and the activation conditions, such as atmosphere composition and temperature. . The presence of inorganic matter, that is, ash components and impregnation agents, which may be detrimental to adsorption or may encourage specific adsorption

To some extent these effects are interactive. For example, inorganic impurities can create disorder within the carbon lattice, resulting in the formation of defects at locations where oxygen can be preferentially adsorbed during activation.

Activated carbons have been divided into two types: H-carbons and L-carbons [ll]. H-carbons are formed at temperatures >7OO0C, typically around l,OOO"C, and are charac- terized by their ability to adsorb hydrogen ions when immersed in water, thereby reducing the pH in bulk solution. L-carbons are activated at temperatures <700"C, usually between 300' C and 4OO0C, and preferentially adsorb hydroxyl ions. Steam-activated carbons are generally used for gold recovery and have predominantly H-carbon type characteristics.

Oxygen is chemisorbed onto carbon more readily than other elements, and the C-0 complexes that are formed can influence surface reactions, wettability, and electrical and catalytic properties of the carbon. Approximately 90% of oxygen on the surface is thought to be present as functional groups. The remainder exists as neutral bonds in ether bridges [12, 131.

A variety of analytical techniques have identified carboxyl, phenol, quinone, and hydroxyl, as well as ester groups (such as lactones, carboxylic anhydride, and cyclic peroxides), at the carbon surface [14, 15, 161. The structures of some of these groups are shown in Figure 7.3. In general, oxide groups formed at low temperature appear to be carboxylates, whereas those formed at higher temperatures tend to be phenolic. The


14

12 n

W

E = a 9 2 2 ; 6

E

c

= 4

2

0 10 100 1,000 10,000 100,000

Pore Radius, R (A) ~~~~

FIGURE 7.2 Pore-size distribution data for a typical thermally activated coal-based activated carbon and a coconut carbon [ l o ]

importance of these groups lies in their ability to affect the acid-base characteristics of the surface. For example, Figure 7.4 illustrates the effect of pH on the zeta potential and acid-base adsorption characteristics of an extruded carbon. This behavior is consistent with the presence of carboxylic acid groups (pK, = 4.8) and phenolic groups (pK, = 9.8), that is, an oxidized surface.

Reduction of surface groups on activated carbon may be represented by the reactions [17] :

Q, + 4H,O + 4e + 2H,Q + 40H- (EQ 7.3)

Q + 2H20 + 2e =t H,Q + 20H- (EQ 7.4)

where H,Q = the quinhydrone group

The potential resulting from Equation (7.4) is given by:

E = Eo(H2Q/Q) - 0.00259 log [HzQ]/[Q] - 0.059 pH + 0.826 (EQ 7.5)

where Eo (HzQ/Q) = 0.699 (V) [211

Most commercial carbons have a reduction potential in the range of 0.1 to 0.4 V (see Figure 4.10).

302 1 THE CHEMISTRY OF GOLD EXTRACTION

0 Q 0

C'

d ! (f)

&-i. c-0

FIGURE 7.3 Structure of surface oxides that have been proposed as being present on the surface of activated carbon: (a) carboxylic acid, (b) phenolic hydroxyl, (c) qulnone-type carbonyl groups, (d) normal lactones, (e) fluorescein-type lactones, (f) carboxyllc acid anhydrides, and (9) cycllc peroxides [ l o ]

400

100

0

I I 0 s -100 us e E

a 8 g. -200

-300 0 2 4 6 8 1 0 1 2

OH

-1 0

-5

0

5

~~

FIGURE 7.4 The adsorption of acid and base by Norit 2020 carbon and the effect of pH on charge and zeta potential [17]


7.1.2 Adsorption from Cyanide Solutions

7.1.2.1 Physical Factors Affecting Adsorption The major physical factors affecting gold adsorption onto carbon are reviewed in the following sections.

Carbons produced using different methods or source materials (Section 7.1.1) have a range of chemical and physical properties, which affect the adsorption rate and loading capacity. In general, higher-activity carbons are softer, due to a more extensive pore structure that reduces the mechanical strength of the carbon. These carbons typically result in higher attrition losses in plants. Attrition losses are important, not only because they consume carbon but also because of the associated gold loss (see Section 7.1.5.8).

The type of carbon required for a particular process application depends on many factors, including the type of adsorption process (i.e., carbon-in-pulp [CIPI, carbon-in- leach [CIL], or carbon-in-columns [CIC]), the gold concentration, solution or slurry flow rate, gold production rate, carbon attrition rate, and the severity of process conditions. High-activity carbons are used when high adsorption efficiency is required, either to pre- vent loss of soluble gold values or to improve overall circuit efficiency; that is, by achiev- ing superior gold loadings or by improving solution equilibria to favor gold dissolution. Lower-activity carbons are used most effectively in circuits that are less susceptible to gold losses resulting from poor carbon adsorption performance and have the advantage of lower attrition losses.

The activity of carbons used for gold extraction decreases with plant usage, and reactivation techniques are commonly employed to limit the extent of this degradation (see Section 7.1.4).

Although the carbon particle size distribution has a significant effect on its external surface area, it has only a very small effect on the specific surface area because of the highly developed internal pore structure. As a result, the ultimate carbon-loading capacity is virtually independent of particle size. However, the size has a large effect on the mean pore length within the carbon particles, and the rate of adsorption increases with decreasing particle size, as illustrated in Figure 7.5. This is an important factor in industrial adsorption systems because the majority of these operate at gold loadings well below the true equilibrium loading capacity of the carbon. Particle size ranges of carbons used in industrial applications typically vary from 1.2 x 2.4 mm to 1.7 x 3.4 mm.

Carbon type.

Carbon partide size.

In practice, several other factors affect the choice of carbon particle size: . The separation of carbon from the solution or slurry phase becomes increasingly difficult at finer sizes (typically, screening of carbon can be performed at 0.7 to 0.8 mm in most slurry applications). Finer carbon is more susceptible to attrition losses because of its higher ratio of surface area to mass, and generally it is reduced to a size where it can leave the plant more quickly than coarse carbon particles. . Smaller carbon has a lower fluidization velocity than coarser carbon, which affects process equipment design (i.e., upflow carbon columns in CIC circuits, carbon elution systems, acid wash vessels, etc.).

Systems that contain carbon with a wide size distribution may experience less of a differ- ence in gold loading with increasing size due to an effect called contact ion exchange [ 191. In this readily measurable effect, gold is transferred from carbon of high gold loading to carbon of low loading-achieved through direct contact of the thin films surrounding the carbon particle, with negligible gold passing into bulk solution [20].


h r

‘L

G 3,000

.I- C m m c I

8 2,000

d a, .I-

1,000

.

’

- \ \

\ - ‘ t

01 ~ ‘ ‘ ‘ ‘ ~ ‘ ‘ ‘ ‘ ~ ‘ ‘ ‘ ‘ i ‘ ‘ ‘ 0.5 1 .o 1.5 2.0

Mean Particle Diameter (mm)

FIGURE 7.5 Rate of gold loading of carbon particles of various sizes [18]

Mixingeflciency. Mixing conditions have an important effect on the gold adsorption rate, as illustrated in Figure 7.6. This effect is due to the fact that most carbon adsorption systems are operated at a pseudo-equilibrium, below the maximum equilibrium loading, where the adsorption rate is dependent to some extent on diffusion through the solid-liquid boundary layer (see Section 7.1.2.3). The pseudo-equilibrium is attributed to the proportion of the pores that are utilized within the operating residence time of the adsorption system. Consequently, the degree of mixing of carbon in a solution or slurry must be sufficient to: . Keep the carbon, solution, and solids suspended and to keep the mixture as

. Maximize the mass transport rate of gold cyanide species to the surface of the car-

Effect ofsolids. The rate of gold cyanide adsorption, decreases with increasing slurry density, as illustrated in Figure 7.7 [18]. This effect is attributed to the following factors: . Decreased mixing efficiency resulting from increased viscosity and decreased

energy input per unit mass of slurry m Physical blinding of the carbon surfaces and pores by fine ore particles . Reduced solution-carbon ratio at higher slurry densities

homogeneous as possible

bon, preferably faster than the actual rate of adsorption at the surface

The mixing efficiency can also be reduced by increased slurry viscosity caused by changes in ore type rather than as a result of a change in slurry density. Ore types that produce high viscosity slurries also have a greater tendency to impair carbon performance; for example, by blinding of carbon pores with very fine particles. This effect is particularly evident when treating clay-bearing ores.

Pulp density is affected by the density of solids, and changes in the type of material treated in adsorption systems must also be considered. For example, when considering


3,000

h - 2,000

v

3 1,000

Carbon Size = <0.70 to >0.50 mrn Solution Conditions:

Ionic Strength = 0 p H = 7 Gold Concentration (pprn):

+ 10 0 100

0 500 1,000 1,500 2,000

Stirring Speed (rpm)

FIGURE 7.6 The influence of Impeller velocity in a baffled reaction vessel on the rate of extraction of gold cyanide by Metsorb 101 actlvated carbon [20]

h - 6.0

- 5.0

Rate Constant

7 60 -

0 10 20 30 40 50

h

'?

E a" v

Percent Solids I I I (masdmass)

10 5 4 3 2 1.5 1

Pulp Water-to-Solids Ratio

FIGURE 7.7 The effect of the pulp solid content on the rate of gold loading onto the carbon [18]

the treatment of calcine (essentially Fe,O,) which has a particle density of 5,000 kg/m3 compared with a quartz-based material with a density of 2,700 kg/m3, the following effects must be taken into account: . Particle sedimentation rates are greater for the higher-density material, and

therefore some settling and dead space may result if inadequate mixing is provided, which reduces solution and carbon mobility.


= The volume proportion of solids is lower for a specific slurry density, and therefore contact between carbon and solution is improved.

7.1.2.2 Mechanism of Gold Adsorption The complex physical and chemical structure of activated carbons allows the adsorption of different species by various mechanisms. Consequently, the exact mechanism of adsorption of gold from cyanide solutions has been difficult to determine; however, during the 1980s a clearer picture emerged.

The adsorption mechanisms proposed prior to about 1978 can be split into four categories, as follows:

9 Adsorption as the Au(1) cyanide ion . Adsorption as molecular AuCN . Reduction and adsorption as metallic gold . Adsorption in association with a metal cation such as Ca2' Studies performed since about 1978 proposed a number of mechanisms that

attempted to account for some well-established adsorption characteristics, most impor- [lo, 21, 22, 231: tantly .

. . . m

.

.

.

Extraction of Au(CN), and Ag(CN), is enhanced by the presence of electrolytes, such as calcium chloride and potassium chloride. Adsorption kinetics and equilibrium loading increase as the pH decreases. The adsorption of gold cyanide increases the pH of the bulk solution. Neutral cyanide complexes, for example, Hg(CN),, adsorb strongly and indepen- dently of ionic strength. Gold cyanide adsorption is a reversible process with generally faster kinetics for desorption under slightly modified conditions. There is some evidence that gold adsorption is dependent on the reduction potential of the system, for example: - Gold adsorption increases with increasing reducing power of the carbon. - Gold adsorption decreases if the carbon has been oxidized by chlorine or nitric

Under most conditions the molar ratio of loaded gold to nitrogen is 0.5:1.0, which is consistent with the presence of the Au(CN), group. Gold adsorption decreases with increasing temperature.

acid.

Detailed investigations using Mossbauer spectroscopy, X-ray photoelectron spectroscopy (XPS or ESCA), and model extractants on high ionic strength solutions, typical of those obtained in actual gold leaching systems, have shown that the gold cyanide complex is adsorbed predominantly as an ion pair [22, 241. Further evidence for this has been provided by surface chemical and other analyses, which have established that the oxidation state of gold on carbon is +l [25]. The mechanism is best illustrated by the equation:

M"' + nAu(CN), --L M"'[Au(CN),I, (EQ 7.6)

where the ion pair, M"'[Au(CN),], is the adsorbed gold species. Detailed experimental evidence leading to this conclusion is available in the literature [lo, 17, 22, 23, 24,251.


10 r

Final Carbon

14,000

n I , b, 5,000

0 6 12 18 24

Time (hr)

Loading (ppm Au)

FiGURE 7.8 An example of kinetics of gold ioadlng onto activated carbon [26]

7.1.2.3 Adsorption Kinetics and Loading Capacity The adsorption of gold cyanide onto activated carbon is dependent upon many chemical and physical factors, which affect both the adsorption kinetics and the equilibrium gold- loading capacity. The initial rate of adsorption of gold cyanide is rapid, with adsorption occurring at the most accessible sites in macropores, and possibly mesopores, but the kinetics decrease as equilibrium is approached (Figure 7.8). Under these conditions the rate is controlled by the mass transport of gold cyanide species to the available activated carbon surfaces. However, once this adsorption capacity has been utilized, a pseudo- equilibrium is established beyond which adsorption must take place in the micropores. This requires diffusion of gold cyanide species along pores within the carbon structure, typically a much slower process than boundary layer diffusion, due to the length and tor- tuosity of the pores [20].

The activation energy for gold adsorption onto carbon has been estimated at 11 kJ/mol, which is well within the range expected for mass transport control [27].

The rate of gold adsorption onto carbon can be described by the first-order rate equation [281:

log C, = mt + log C, (EQ 7.7)

where C, = gold concentration at time, t C, = initial gold concentration m = a rate constant which can be readily determined from a plot of log C versus

time, using data obtained from simple laboratory tests

A typical equilibrium gold-loading isotherm is given in Figure 7.9. The loading capacity of carbon has traditionally been expressed as an iodine number (the mass of iodine adsorbed per gram of carbon in a 0.02 N iodine solution) or as a carbon tetrachlo- ride number (weight percent CCl, loading on carbon exposed to air saturated with CC1, at 0°C) . Both of these values provide a useful approximation of the available surface


30,000

h 10,000

k

s s

Q

c 0

v

e

0 1,000 0

c

100 .

-

-

. - Freundlich Equation

or log Q= const + n log C

- Freundlich Equation

or log Q= const + n log C

I I I I I 0.01 0.10 1 .o 10.0

Gold in Solution (mg/L)

FIGURE 7.9 Equilibrium adsorption isotherm for loading of gold on carbon 1261

area for some vapor phase adsorbates, but it has been demonstrated that such estimates are poorly correlated with gold adsorption capacity [291, due to the complex combina- tion of physical and chemical processes involved in gold adsorption from cyanide solutions. As a consequence, actual gold-loading rate data are generally of more practical use for optimizing industrial adsorption systems, particularly since true equilibrium between gold in solution and gold on carbon is never attained.

For the same reason, it is most appropriate to use an empirically developed equilibrium gold-loading capacity (K value) for the evaluation of carbons for use in gold adsorption systems. This is obtained by reacting various weights of carbon with a standard borate- buffered gold solution for a fixed time. The results are plotted as the Freundlich isotherm (Figure 7.9), and the K value is interpolated as the carbon loading in equilibrium with a residual gold solution concentration of 1 mg/L [291.

Several variations of these expressions for loading rate and capacity have been developed and applied for specific operations around the world, and further information is available in the literature [18, 28, 29, 30, 311.

7.1.2.4 Chemical Factors Affecting Adsorption Efficiency The major chemical factors affecting the efficiency of gold adsorption onto carbon are reviewed in the following sections.

The adsorption of gold onto carbon is exothermic, which accounts for the ability to reverse adsorption by increasing temperature [lo]. Consequently, the loading capacity decreases as the temperature increases, as shown in Figure 7.10 and Table 7.1. This is exploited in the high-temperature elution of gold from loaded carbon, discussed in detail in Section 7.1.3. The adsorption rate increases slightly with increasing temperature (see Table 7.1 and Figure 7.11) due to the accelerated diffusion of gold cyanide species, following a behavior described by the Arrhenius equation (4.39).

The rate of gold adsorption and the equilibrium loading capacity both increase with increasing gold concentration in solution, as illustrated in Figure 7.9. Typically gold-loading rates of 10 to 100 g Au/hr/t carbon and loadings of 5 to 10 kg Au/t carbon are achieved in practice at gold concentrations produced by standard cyanide leaching processes (Chapter 6).

Temperature.

Gold concentration in solution.

[CH. 7 SEC. 7.11 SOLUTION PURIFICATION AND CONCENTRATION I 309 400

350

300 h

p1 1

250 v

C 0 5 200 0

150 z

100 s

50

0 0.0 0.2 0.4 0.6 0.8 1 .o 1.2

Equilibrium Gold Concentration (mmol/L)

FIGURE 7.10 Equilibrium adsorption isotherms for gold cyanide on carbon at dlfferent temperatures (experimental condltions: volume of solutlon = 50 mL; mass of carbon = 0.25 g; adsorption medlum contained 2.8 g/L CaCI, and 0.5 g/L KCN) [31]

TABLE 7.1 Effect of temperature and sodium cyanide concentratlon on gold loading: [Au] = 25 mg/L, pH 10.4 to 10.8 [20]

Temperature Free Cyanide Rate Constant, k Gold-Loading Capacity (“C) (mg/L) (per hr) (mglL) 20 0 3,400 73,000 25 130 3,390 62,000 24 260 2,620 57,000 23 1,300 2,950 59,000

44 43 42 43

62 62 62 62

0 130 260

1,300

0 130 260

1.300

4,190 4,070 3,150 3,010

4,900 4,920 3,900 4,060

48,000 47,000 42,000 33,000

35,000 29,000 29,000 26,000

81.5 260 5,330 20,000

Cyanide concentration. Both the loading rate and capacity of gold on carbon decrease with increasing free cyanide concentration. This is illustrated in Table 7.1, which shows data for tests at constant ionic strength, and the effect is attributed to increased competition of free cyanide species for adsorption sites on the carbon [20]. However, the selectivity of activated carbon for gold over other metal cyanide species increases with increasing cyanide concentration, as exploited in the treatment of high copper ores (see Section 7.1.2.5).


Conditions: Ionic Strength = 0.1 M

0

0 c

0 1

a, m cc c

I

2.8 2.9 3.0 3.1 3.2 3.3 3.4

1/T, A-' x lo3

FIGURE 7.11 Effect of temperature on the rate of gold extraction [20]

In practice, the cyanide concentration used in adsorption systems is often determined by the requirements for optimal gold dissolution and by natural cyanide degradation rates within the extraction circuit (i.e., 0.1 to 0.3 g/L sodium cyanide [NaCNl).

A decrease in solution pH increases both the adsorption rate and loading capacity, as shown in Figure 7.12 and Table 7.2. The effect on the adsorption rate is quite small over the pH range 9 to 11, as applied in cyanidation circuits, with only a small advantage to be gained by reducing pH. The capacity is increased by approximately 10% as the pH is lowered from 11 to 9. In practice, the pH is usually maintained at >lo to avoid loss of cyanide by hydrolysis, or alternatively the pH may be allowed to decrease naturally through a CIP or CIL circuit to assist with cyanide degradation prior to tailings disposal.

Ionic strength. The effect of ionic strength on gold adsorption is shown in Table 7.2. This effect is also illustrated by the finding that the gold cyanide complex can be eluted off carbon with deionized water. Both the adsorption rate and loading capacity are increased with increasing ionic strength [32].

Under laboratory conditions, gold loading capacity increases with increasing concentration of cations in solution in the following order:

SolutionpH.

Concentration of other metab.

Ca2' > Mg2' > H' > Li' > Na' > K+

and decreases with anion concentration in the order:

CN- > S2- > SCN- > S2032- > OH-> C1- z NO,

These effects are compounded under industrial conditions by the adsorption of other metal cyanide species, which compete for active, available adsorption sites. This results in a slower adsorption kinetics and reduces the equilibrium capacity for gold. The adsorption of these metals is considered further in Section 7.1.2.5.

The beneficial effect of oxygen on the adsorption of gold from cyanide solutions has been reported [33]; however, the effect, as has been demonstrated, is most significant in low ionic strength solutions, which are atypical of most industrial leach solutions. Despite this, some benefit is observed in actual adsorption

Dissolved oxygen.


40 -

I I I I I I I I I I I 1

FIGURE 7.12 Effect of pH value of the adsorption medium on the gold capacity of the carbon; experimental conditions: volume of solution 300 mL, mass of carbon = 0.25 g, nltrogen atmosphere, initlal concentration of gold = 190 mg/L 1231

TABLE 7.2 Effect of pH and ionic strength on rate of loadlng and equilibrium capaclty: ionic strength = 0.2 M; pH 6.5 [20]

Gold-Loading Rate Constant, k Capacity

vH (ver hr) (rna/L)

Ionic Strength

(M) Rate Constant, k

h e r hr)

Gold-Loading Capacity (rnalL)

1 1.3 3,010 45,000 9.1 3,000 86,000 7.1 3,660 92,000 4.2 3,900 122,000 3.1 4,420 143,000 1.5 4,880 2 16,000

0.005 0.010 0.020 0.050 0.100 1 .ooo

3,150 3,690 3,480 3,902 3,310 4,150

56,000 60,000 63,000 73,000 84,000

11 3,000

systems-attributed to the catalytic oxidation of cyanide. This results in a decrease in cyanide concentration which favors adsorption.

Carbon fouling, or poisoning, due to the adsorption, precipitation, or physical trapping of other solution species and ore constituents can have a severe adverse effect on gold adsorption efficiency, as considered in Sections 7.1.4.1 and 7.1.4.3.

Carbon fouling.

7.1.2.5 Adsorption of Other Metals Leach solutions usually contain a variety of metal ions and complexes, including silver, copper, nickel, zinc, iron, and mercury, which are adsorbed onto activated carbon to varying extents, depending on the concentration of each species, the properties of the carbon, and the solution conditions. The adsorption of silver, and in some cases mercury, may be important as they may be economic by-products of gold. In contrast, the adsorption of noneconomic metals is detrimental to gold extraction, because these species compete with gold (and silver) for active carbon sites. In addition, the adsorbed metals may be difficult to desorb under the conditions most suitable for gold desorption, resulting in a buildup of the metals on the carbon, which decreases carbon activity. Any metals


adsorbed, for example, copper or mercury, may contaminate the final product, requiring additional treatment step(s).

Fortunately, activated carbon is highly selective for gold and silver over most other metal species, mercury being the most important exception. The general order of preference of adsorption for several commonly encountered metal complexes is given as follows:

Au(CN), > Hg(CN), > Ag(CN), > Cu(CN)i- > Zn(CN)$ > Ni(CN):- >> Fe(CN)i-

High loadings of nonprecious metals can be achieved onto activated carbon in the absence of significant precious metal values, which makes it possible for heavy metal ions to be removed from water, a process commonly used for water purification.

The mechanism of Ag(1) cyanide adsorption is similar to that of Au(1) cyanide; however, the adsorption capacity of carbon for silver is substantially less than that for gold [lo]. Also, the gold complex tends to displace silver from carbon. Both these factors are important in plant design and operation because a larger amount of carbon is required to recover an equivalent amount of silver from solution, and the silver loading rate is slower than gold. Many gold ores contain significant quantities of silver, often many multiples of the gold concentration, potentially making them less amenable to treatment by carbon adsorption (see also Sections 3.3.5 and 12.2.3.1).

The neutral mercury cyanide complex, Hg(CN),, competes directly with Au(CN), for adsorption sites and can even displace some of the adsorbed gold from carbon. Fortunately, mercury is usually present in leach solutions in relatively low concentrations, partly due to its low grade in most ores and its rather poor dissolution characteristics, and it rarely has a severely detrimental effect on gold adsorption. However, the highly effective adsorption and desorption of mercury under conditions applied for gold extraction require that, for treatment of ores containing significant quantities of mercury, a method of removing mercury must be provided downstream (see Chapter 10).

The adsorption of copper is strongly related to pH and cyanide concentration. The Cu(CN), complex, which is favored at low pH and low cyanide concentrations, is most readily adsorbed, whereas at high pH and high free cyanide concentration, the Cu(CN):- complex predominates, and adsorption is poor (see Figure 6.18). Therefore, the adsorption of copper species increases in the order:

Cu(CN):- < Cu(CN)i- < Cu(CN),

The detrimental effect of copper on gold adsorption has been reported extensively, and copper concentrations as low as 100 mg/L can interfere severely with adsorption processes [34]. The effect of cyanide concentration on gold- and copper-loading capacities is shown in Figure 7.13. In order to minimize the adsorption of copper onto carbon (and to minimize the adverse impact on gold loading), the molar ratio of CN-Cu should be maintained at or above 4: 1 in leach solutions prior to feeding the carbon adsorption process. Alternatively, copper can be allowed to co-adsorb onto carbon with gold and can be removed selectively using a cold desorption step (see Section 7.1.3.8).

Processes that treat materials containing high concentrations of cyanide-soluble copper, that is, yielding >200 mg/L Cu in solution, will require very careful control of pH and cyanide to allow satisfactory treatment. In the extreme case, these materials may be unsuitable for treatment by carbon adsorption [351.

Silver.

Mercury.

Copper.

7.1.3 Elution

Activated carbon that has been loaded with gold and other metals in adsorption processes must be treated by an elution step to desorb the metals from the carbon. This produces a

[CH. 7 SEC. 7.11 SOLUTION PURIFICATION AND CONCENTRATION I 313 Conditions:

Au in feed = 9 pprn Cu in feed = 25 pprn

0 2 4 6 8 10

Cyanide-Copper Ratio

FIGURE 7.13 Effect of the ratio of cyanide to copper on the relative equilibrium loading capacities of copper cyanide and gold cyanide [20]

smaller volume of high-grade gold solution, suitable for final gold recovery by electrowinning or zinc precipitation (Chapter 8), and allows the carbon to be recycled to the adsorption circuit. Carbon is typically reused between 100 and 400 adsorption-elution cycles, depending on the carbon quality and the effectiveness of reactivation procedures applied (see Section 7.1.4).

The desorption process, commonly referred to as either elution or stripping, is a reversal of the adsorption process, and the chemical and physical factors that inhibit adsorption generally enhance desorption. For gold adsorbed from cyanide solutions, the desorption reaction is most simply represented by:

Mn'[Au(CN)Jn(ads) + n Au(CN), + Mn'

although the exact mechanism of adsorption is considered to be more complicated.

(EQ 7.8)

7.1.3.1 Temperature and Pressure Temperature is the most important factor in the elution of gold cyanide from carbon, with approximately an order of magnitude increase in the elution rate for a 100°C increase. For example, the elution rate at 180°C is eight times faster than at 90°C at atmospheric pressure. Figure 7.14 shows the effect of temperature on gold desorption efficiency, given for the example of the Zadra elution scheme. Although it is possible to reduce elution times substantially by operating at temperatures >looo C, this requires the use of elevated pressure to keep the eluting media in the liquid phase and enable practical application of the system. Consequently, elution systems have evolved into two classes: . Processes that operate at atmospheric pressure and temperatures < 100°C

Processes that operate at elevated pressures to allow operation at elevated temperatures, that is, >1OO"C, to achieve faster elution rates


90

80

70 h

8 60

50

0 .- I

30

20

10

0

Strip Solution (bv)

FIGURE 7.14 The effect of temperature on gold desorption employlng Zadra process solutlon [36]

Unpressurized systems operate at temperatures just below the solution boiling point (95"C), whereas pressurized systems have been operated as high as 160°C and at 500 kPa. At temperatures >180°C, most metal cyanide complexes, including gold cyanide, decompose to the metallic species and free cyanide, which results in high residual gold concentrations on carbon, which are very hard to remove [37]. Cyanide decomposition, resulting in ammonia evolution, also increases at elevated temperatures (see Chapter 6).

7.1.3.2 Cyanide Concentratlon The effect of cyanide on the rate of gold desorption is illustrated in Figure 7.15. Increas- ing cyanide concentration increases the competition of cyanide ions with gold cyanide species for adsorption sites on the carbon and assists with the displacement of gold cyanide species from the carbon. However, the presence of free cyanide throughout the desorption process is not a requirement for effective elution (as illustrated by the OH- line in Figure 7.15) and several procedures have been developed that use a cyanide presoak step followed by deionized water elution. Consequently, elution systems can be divided into those using cyanide throughout the process and those using cyanide during a presoak only, as described in Section 7.1.5.6.

7.1.3.3 Ionic Strength Ionic strength has a greater effect on elution rate than cyanide concentration, as shown in Figure 7.16. Gold may be desorbed quite effectively with low ionic strength solution, for example, deionized water, even in the absence of free cyanide 1381. The beneficial effect of divalent cations, such as calcium and magnesium, on gold adsorption onto activated carbon is reversed for elution processes.


I -

I /

OH-

0 0.2 0.4 0.6 0.8 1.0 1.2

Concentration (mollL) ~ ~ ~~

FIGURE 7.15 Varlatlon of the rate of elution with eluant strength at a constant lonlc strength of 1.2 at 95°C [38]

7.1.3.4 pH Elution processes that have been developed since the original Zadra process (1950) have all used solutions containing 1% to 3% sodium hydroxide, either in a presoak step or for the main elution step [39, 401. The hydroxide ions displace gold cyanide ions on the carbon in a manner similar to free cyanide ions, as discussed in Section 7.1.3.2, and as illustrated in Figure 7.15. In addition, sufficient alkalinity is required to avoid loss of cyanide by hydrolysis, that is, to typically maintain the pH between 10.0 and 12.0. The control of pH during elution is most important for processes that use zinc precipitation for subsequent gold recovery, because insoluble zinc hydroxide can be formed (Chapter 8).

7.1.3.5 Organic Solvents The rate of elution can be significantly increased by the addition of organic solvents, such as alcohols and glycols, to the aqueous eluant. These increase the activity of other ionic species in solution, particularly smaller ions such as cyanide, in preference to larger ones, for example, gold cyanide. This effect increases the efficiency of displacement of gold cyanide from carbon.

A variety of solvents have been tested and used to enhance elution processes over a range of temperatures (Figures 7.17 and 7.18). Alcohols, such as ethanol, methanol, and isopropanol, at concentrations of 15% to 25% in the eluant, can be used to reduce elution times by a factor of 3 to 4 (e.g., from 48 to 12 hr for Zadra-type elution at 90°C). Unfortunately, these solutions are highly flammable and may constitute a fire hazard in industrial plants. Glycols, such as ethylene and propylene glycol, have been used in similar concentrations but with generally smaller reductions in elution times-approximately half the reductions that are achieved with alcohols. Glycols are less flammable than alcohols and therefore may be preferred in some cases.

The relative effectiveness of various organic solvents decreases in the order [41] :

acetonitrile > methyl ethyl ketone acetone >>> demethyl formamide > ethanol


0.075

h r

'L

's 6 0.050

e E P 9

c v

0

8

8 c

c 0 c 0 .- -

0.025 c 0 a m K .I-

0

0.2 M CN- 0.2 M OH-

0.2 M CN- 0 M OH-

0.2 M OH- 0 M CN-

0 0.2 0.4 0.6 0.8 1.0 1.2

Ionic Strength (mol/L)

FIGURE 7.16 Effect of the ionic strength of the eluant on the rate of elutlon at 95°C [38]

In addition, the use of organic solvents in eluant solutions without cyanide has been proposed. Solutions containing 1% by weight of sodium hydroxide with 20% (by volume) of a suitable organic solvent, such as isopropyl alcohol, ethylene glycol, or ethanol, have been demonstrated to achieve effective elution at 8 0 ° C in about 8 hr [421. The effectiveness of the various solvents decreases in the following order:

isopropyl alcohol > ethylene glycol > ethanol

The use of organic solvents at high temperatures may benefit the removal of adsorbed organic species (e.g., oils, humic acid, etc.) from carbon [431. This does not appear to have any significant adverse effect on subsequent electrowinning or zinc precipitation recovery processes.

7.1.3.6 Solution Flow Rate The solution flow rate applied for elution is usually expressed as carbon bed volumes per hour (bv/hr). The elution rate tends to be virtually independent of flow rate above about 1 bv/hr, but the residual gold loading on carbon decreases with increasing flow rate after a fixed time, as shown in Figure 7.19. Typically, flow rates of 2 to 4 bv/hr are used to produce a carbon with low residual gold loading, while avoiding the need to treat exces- sively large volumes of eluant.

[CH. 7 SEC. 7.11 SOLUTION PURIFICATION AND CONCENTRATION I 3 1 7

- 0 5 10 15

100

90

80

70

h

60

' 50

C 0 .- I

8 8 $ 40

30

20

10

n 240 20

0

30

60 h

5 v

90 5 8 5 120 [5)

e

C C a .- .-

150 5 n s

180

210

FIGURE 7.17 Effect of ethanol addltlve to Zadra solution on gold desorptlon [36]

(Gold on Carbon = 3,800 pprn; 40% v/v Solvent with NaCN at 10 g/L)

0.8 C 0 .- I

P 8 0.6 8 - a 5 0.4

s .- 1

LL 0.2

0

Acetonitrile Acetone

3 Dernethyl Forrnarnide

D----.--il. Ethanol

I I J 0 10 20 30 40 50

Time (hr)

FIGURE 7.18 Desorption of gold by organic solvents at 25°C [23]


5,000

4,000 h

5 OI

6 3,000

1,000

0

Flow Rate (bv/hr)

FIGURE 7.19 Effect of flow rate on average loadlng of resldual gold after 20 hr [38]

7.1.3.7 Gold Concentration in Solution The elution rate decreases, and the residual gold loading increases, with an increase in the gold concentration of the eluant (Figure 7.20). This reduces the rate of elution with time in a batch process-a most important factor for elution systems that recirculate solutions directly following metal recovery. An example of this is atmospheric Zadra elution coupled with electrowinning or zinc precipitation for metal recovery. In this case, the gold concentration in the solution used for elution, and consequently the efficiency of elution, depends on the efficiency of the associated electrowinning or zinc precipitation step.

7.1.3.8 Elution of Other Metals Metal cyanides, other than gold, are also eluted from carbon under the conditions applied for gold desorption. Copper, silver, and mercury preferentially desorb before gold, as illustrated for copper and silver in Figure 7.21. The possibility of sequentially desorbing copper and then precious metals in a two-stage elution process has been applied by operations that experience high copper loadings on carbon, for example, El Indio (Chile) (see Section 12.2.6.3). In this case, copper (and other base metal cyanides) are eluted from loaded carbon using a cold (ambient temperature), alkaline cyanide solution. Depending on the metal loading on carbon and the eluant properties, >90% of the copper (and potentially other base metals) can be removed selectively over gold and silver. The gold and silver (and a portion of the remaining base metal complexes) are subsequently desorbed using one of the elution processes described in Section 7.1.5.6.

7.1.4 Carbon Fouling and Reactivation

Carbon fouling is the buildup of organic and inorganic substances on carbon, which detri- mentally affect gold adsorption. This results in a decrease in the kinetics and equilibrium


1,200

1,000

0 800 cn C v

: 0

5 600 2 s - m 3

u) 2

$ 400

200

0 0 5 10 15 20 25

Gold in Solution (g/t)

FIGURE 7.20 Equilibrium isotherm for the distribution of aurocyanide between activated carbon and a solution containing 0.2 M NaOH and 0.2 M NaCN at 95°C [38]

loading of gold adsorption onto carbon and may adversely affect the efficiency of desorption processes by reducing desorption kinetics and elevating residual (eluted) carbon gold loading. Fouling may also affect the eluate composition adversely, for example, by the formation of silica gel or release of particulate matter into the eluate solution.

The effectiveness of carbon adsorption as a commercial process relies on the ability of activated carbon to be reused many times, which depends on the degree of fouling and the efficiency of any reactivation processes used. Carbon fouling can occur by any or all of the following mechanisms:

= Undesirable organic or inorganic species are adsorbed onto the carbon surface, taking up active sites which would otherwise be available for gold adsorption . Inorganic salts are precipitated onto the carbon surface, blocking active sites . Solid particles or precipitates are physically trapped in carbon pores, restricting access to gold-bearing solution

In industrial systems this fouling can be counteracted in three ways: . Reducing fouling during adsorption and desorption processes


5 1,500

t 0 Gold

W Copper ,000 + Silver

500

0 0 20 40 60

Strip Solution Volume (m3)

FIGURE 7.21 Typical elution curves from carbon stripping [44]

. Removing the fouling species, following adsorption or desorption Removing fouled carbon from the circuit and replacing with new carbon

Consequently, it is important to understand the sources and nature of carbon fouling, as well as the methods available for reactivation, in order to optimize carbon adsorption systems. These are reviewed in more detail in the following sections.

7.1.4.1 Inorganic Fouling The most important forms of inorganic fouling are [45] :

Calcium salts, primarily carbonate, but also to a lesser extent sulfate and other species . Magnesium and sodium salts . Fine ore minerals, such as silica, complex silicates, and aluminates (including

. Fine iron particles and associated products resulting from grinding media clays and clay-forming minerals)

Base metal precipitates from the leach solution The mechanisms by which inorganic salts are deposited onto activated carbon are largely unrelated to the adsorption of gold species.

Calcium carbonate is formed by carbon dioxide from the atmosphere dissolving in water to form C0:- which reacts with available Ca2' ions as follows:

CO, + H,O --L HCO; + H'

HCO; + H' + CO$

(EQ 7.9)

(EQ 7.10)

[CH. 7 SEC. 7.11 SOLUTION PURIFICATION AND CONCENTRATION 1 321

~ 0 3 2 - + Ca2+ + C ~ C O ,

The most common sources of calcium ions in gold extraction are from lime (CaO) and slaked lime (Ca(OH),), added to process slurries and solution for pH control, and soluble ore constituents (i.e., limestone or dolomite).

Alternatively, carbonate ions may be formed by the oxidation of cyanide at the carbon surface:

2CN- + 0, + 4H,O --L 2CO;- + 2NHi (EQ 7.12)

Ca2+ + ~ 0 3 2 - --L CaCO, (EQ 7.13)

At higher temperatures, the formation of methanoic and oxalic acids is favored. There is evidence that the carbon surface imposes a region of solution which is atypical of the bulk solution and in which the local solubility of metal compounds is reduced. This leads to a precipitation reaction on the carbon surface. This behavior is similar to the adsorption of polyvalent metal species onto oxide and silicate minerals [47]. The solution conditions for metal deposition are similar to those for precipitation in bulk solution; however, adsorption occurs at lower pH values and lower metal ion concentrations [46, 481. Although hydroxides may be precipitated onto carbon, carbonate precipitation is usually the more important problem in industrial systems.

Studies of calcium deposits using scanning electron microscopy (SEM) have revealed the material to be crystalline 1451. Such precipitates fill the cracks and depressions on the external surfaces of the carbon, and are widespread within the pore structure, even after acid washing. This inhibits the diffusion of gold cyanide species through carbon pores, thus reducing both the active surface area and the adsorptive properties of the carbon.

The precipitation of calcium carbonate can largely be avoided by reducing the pH to <8.3, below which the more soluble calcium bicarbonate is formed. However, in most cyanide adsorption systems this is impractical, and possibly hazardous, because of the increased hydrogen cyanide evolution. On the other hand, a possible environmental advantage of such practice is the associated destruction of free cyanide in CIP and CIL circuits prior to discharging slurry tailings.

Descalant reagents can be added to process solutions to reduce calcium carbonate precipitation on carbon in CIC systems. These have been used successfully at many operations, for example, Chimney Creek, Carlin, and Pinson (Nevada, United States).

7.1.4.2 Inorganic Removal Many inorganic foulants can be removed by acid washing, whereby the precipitated salts are dissolved in dilute mineral acid (HC1 or HNO,) and then rinsed from the carbon. Clearly, the success of this technique depends on the solubility of the salt deposited, and the type and concentration of acid solution used. Dilute mineral acids will readily dis- solve calcium carbonate and many other metal salts, leaving the adsorbed gold species essentially unaffected. Consequently, acid washing may be performed either before or after the desorption (elution) of precious metals from the carbon. Several factors affect this decision, discussed in Section 7.1.5.6.

The general equation for the dissolution of a divalent metal carbonate in mineral acid is given as follows:

MCO, + 2H' =' M2' + CO, + H,O (EQ 7.14)

In practice, both hydrochloric and nitric acid have been used for acid washing.


Hydrochloric acid. The preferred reagent in industry, hydrochloric acid has been applied at both ambient and elevated temperatures, up to approximately 85°C. Volume concentrations of 1% to 5% HC1 are used, depending on the loading of inorganic foulants on the carbon and the acid washing conditions applied. The efficiency of calcium removal is largely unaffected by acid concentration (3% to 10% HC1) and temperature (25°C to 90°C) for calcium loadings below approximately 1% Ca. However, significant benefits may be gained by increasing acid concentration (i.e., to 5%-7.5%) for carbon loaded with higher calcium levels, that is, >2% Ca (see Figure 7.22).

The efficiency of calcium removal is strongly related to the efficiency of carbon-acid contact (mixing) during acid washing, as is the case for other carbon adsorption and desorption processes. Because most carbon acid washing systems are operated as stationary, semifluidized, or fully fluidized beds, the calcium removal efficiency depends on the vessel geometry and the acid flow rate. Also, the dissolution rate of calcium salts within carbon pores is diffusion controlled. Consequently, the process residence time is likely to be important in optimizing removal of inorganic salts. An example of the effect of acid strength and reaction time on final calcium loading is shown in Figure 7.22.

In well-mixed systems, with relatively low calcium loadings ( ~ V O , for example), residence times of 10 to 15 min may be adequate, whereas for poorly mixed systems or when high calcium loadings are experienced (i.e., >3%) over an hour may be required.

Dilute hydrochloric acid is usually capable of removing between 80% and 95% of calcium loaded on carbon. Although the efficiency of removal of sodium and magnesium salts may be somewhat lower (c80%), these species are usually present on carbon in lesser quantities than are calcium salts, and the efficiency of their removal is generally not as critical. In addition, up to 50% of nickel, zinc, iron, and silicon loaded on the carbon may also be removed, depending on the adsorbate and the acid washing conditions applied. Silver, mercury, and copper are not removed from carbon by dilute hydrochloric acid.

The major drawback associated with the use of hydrochloric acid is the presence of residual chloride ions in carbon pores, which are difficult to remove effectively following acid washing. This may cause severe problems when acid washing precedes thermal reactivation processes because the highly corrosive chloride ions are released and vapor- ized at elevated temperatures. Similarly, if acid washing precedes desorption (elution), residual chloride ions may be released, creating a corrosive environment in the elution circuit, although this is generally less severe and can often be controlled. This effect, and the possible reprecipitation of salts, is controlled by washing and neutralizing acid- washed carbon with sequential water and sodium hydroxide washes. Despite this, 100% removal of chloride species is rarely achieved. Alternatively, acid washing may be performed after desorption and thermal reactivation to allow chloride release to occur after the carbon has been returned to the adsorption circuit, as discussed in Section 7.1.5.6.

Used in some operations, particularly in North America, nitric acid is applied in a similar manner to hydrochloric acid. The use of nitric acid avoids the corro- sion problems associated with the use of hydrochloric acid but may cause other problems, such as oxidation and deactivation of carbon surfaces. This effect is thought to be small in very dilute solutions ( ~ 5 % HNO,) but increases with increasing concentration. Mercury, and to a lesser extent silver, are removed from carbon by dilute nitric acid, which may be beneficial or detrimental, depending on their concentrations and the specific process requirements. Finally, the use of nitric acid introduces nitrate ions into the process which may, depending on the nature of the process, present environmental problems.

Nitric acid.


c 0

% 0 c 0

2.5

2.0

1.5

1 .o

0.5

Strength 3.3%

-

a c 0 0.5 1 .o 1.5 2.0

Time of Acid Treatment (hr)

NOTE: Percentages indicate HCI starting concentration during treatment.

FIGURE 7.22 Effect of acld strength and tlme on the residual calcium content of carbon at Grootvlei (South Africa) 1491

7.1.4.3 Organic Fouling Because activated carbon is a relatively nonpolar and hydrophobic material, it readily adsorbs most organic compounds from aqueous solutions. The organics that contribute most to fouling are as follows: . Diesel oil, lubricating oils, and antifreeze chemicals from mining and processing

equipment . Humic acid and other vegetation decomposition products that are present in the ore . Flotation reagents, such as collectors and frothers . Flocculants and other surface-active reagents Adsorption occurs by mechanisms involving hydrogen bonding and van der Waals attrac- tion between organic species and the carbon surface. It is most favorable under conditions of limited solubility, for example, when the potential adsorbate has high molecular mass, low polarity, or low ionization potential.

Such fouling can lead to a significant proportion of the carbon surface area being inaccessible to gold cyanide species, resulting in decreased adsorption rate and decreased gold loading. This principle is exploited when blocking agents are employed to deactivate organic carbon components in carbonaceous ores (Section 5.6).

7.1.4.4 Organic Removal Organic species that foul activated carbon fall into two categories with respect to their removal [SO] :


Adsorbates that are either highly volatile or easily thermally decomposed to gas- eous products at normal kiln temperatures (500°C to 800°C). . Nonvolatile species that leave a carbonaceous residue on pyrolysis. These residues can be removed using steam at temperatures >650°C, according to the following reaction:

(C), + nH,O(steam) + nCO + nH, (EQ 7.15)

This reaction also causes some loss of the original carbon, particularly if inorganic salts of calcium, magnesium, and iron are present, which catalyze the carbon- steam reaction and should therefore be removed by acid washing prior to thermal reactivation (kilning).

Fouled carbon can be regenerated by heating to 650°C to 750°C in a nonoxidizing atmosphere. A steam atmosphere is often used for the reason previously discussed and as illustrated in Table 7.3. The most important variables during thermal reactivation are the following: . Temperature . Steam addition . Residence time . Initial moisture content of carbon . Presence of extraneous mineral matter . Reactivation equipment type Figure 7.23 (a and b) shows the effects of temperature, retention time, and steam addition on relative carbon activity for a South African gold plant [501. In this case, it was also observed that carbon loss was reduced and activity was increased by acid washing prior to the elution step.

If the temperature or residence time is too low, then removal of the organic material may be incomplete. If the temperature is too high, then further activation of the carbon may occur, resulting in increased material loss and decreased hardness. The latter is due to an excessive increase in pore structure, which reduces mechanical strength. To reduce energy consumption, the carbon should be dewatered prior to reactivation.

Hot, regenerated carbon should be cooled by quenching in water, which minimizes exposure to oxygen and maintains activity. Some operations use warm water for quenching to avoid excessive thermal shock on the carbon, which might otherwise result in particle fracturing and degradation.

The performance of kilning may also be affected by the presence of coarse mineral particles, wood chips, and plastics (e.g., electrical cable insulation). Mineral particles may be removed by acid washing, then by separation on a jig or shaking table if necessary. Kiln off-gases consist primarily of carbon monoxide, carbon dioxide, hydrogen, and steam, Small quantities of other substances (e.g., hydrogen sulfide, carbon disulfide, and ammonia) may result from decomposition of flotation reagents and adsorbed cyanide. Any residual mercury on the carbon is readily volatilized during thermal reactivation, and the effects of mercury should be carefully considered on a case-by-case basis.

7.1.5 Process Considerations

7.1.5.1 Carbon Preparation Despite being manufactured to particular size specifications, fresh carbon always con- tains small quantities of fine carbon, which is unsuitable for use in carbon adsorption


TABLE 7.3 The effect of steam flow rate and contact tlme on the activity of carbon using Rintoul thermal reactlvatlon kiln [18]

Percentage Activity Compared to Fresh Carbon (kglhr) After 1 hr After 15 hr 125 77.9 83.5

Steam Flow Rate

150 91.5 94.0 175 93.9 99.4

systems. In addition, much of the carbon produced by manufacturers is angular in shape, containing points and sharp edges which are readily attrited during transportation and handling under normal process conditions. Similarly, flat (plate-like) particles and damaged (i.e., cracked or fractured) particles are susceptible to rapid degradation in adsorption systems, resulting in the loss of carbon fines and any contained gold.

Consequently, it is desirable to remove the carbon fines that are most rapidly gener- ated from fresh carbon prior to introducing it into the adsorption circuit. This is usually achieved by vigorous mechanical agitation in water at 10% to 20% solids for 0.5 to 2 hr. Typically, between 1% and 3% of the total carbon weight is removed as fines, depending on the carbon type and the severity and duration of attritioning. The conditioned mixture (coarse carbon, carbon fines, and water) is then screened at a size slightly coarser than the screen size employed for interstage screening within the adsorption circuit. The coarse carbon produced is ready for use, while the fine carbon slurry is discarded from the process.

7.1.5.2 Carbon-In-Pulp Process A well-established technology, the CIP process, is commonly applied for the extraction of gold from cyanide leach slurries. The process is usually configured with carbon flowing counter-current to the process slurry in mechanically agitated tanks (Figure 7.24). An alternative to this is a carousel-type system, which will be discussed later. The slurry is introduced to carbon following or during cyanide leaching and passes through several stages of carbon adsorption, the number depending principally on the tank sizes, the carbon concentration, and the amount of gold to be adsorbed. The number of stages ranges from four to ten, with five to six most typical. The gold concentration in solution is depleted as the gold is adsorbed onto carbon. Fresh or reactivated carbon is introduced at the tail end of the process and is transferred either continuously or in batches up the adsorption stages, in the opposite direction to the slurry although at a much slower rate. The carbon in each stage becomes loaded to the pseudo-equilibrium, which depends on the gold concentration in solution in each stage. The carbon in the first stage has the highest gold loading and is contacted with the highest-grade solution while the carbon in the last stage has the lowest loading and, consequently, the highest activity.

Carbon is retained in each adsorption stage by interstage screens with apertures slightly smaller than that of the activated carbon, which are large enough to allow free flow of slurry between the stages. The efficiency of interstage screening has proven to be one of the biggest challenges in the application of carbon adsorption and has received much attention. The major screen types and their approximate slurry-handling capacities are summarized as follows 1521 : . Horizontal screen with wiper (e.g., Kambalda): 40 to 60 m3/hr/m2

Vibrating screen with inclined deck (e.g., Derrick): 90 to 110 m3/hr/m2

- Vertical, cylindrical, with wiper and/or bottom discharge (e.g., NKM, Kemix): 30 Equal pressure, air-cleaned (EPAC): 10 to 50 m3/hr/m2

to 60 m3/hr/m2


20

10

0

(a) 100 r Steam Flow Rate (0.5 kg/hr)

- 7 No Steam; 10 min at Temperature 0 No Steam; Up to Temperature and Discharged

0 Eluted Carbon -

I I I I I 1

"t 70

- + Total Pore + 0 Micropore v Mesopore

- A Macropore

+ Virgin Carbon

0.600

- a, 0.500 0 r m .- a 5 0.400

k a

m

6

v 6 0.300

0

0 Eluted 2 hr 4 hr 8 hr Virgin

Carbon Sample

FIGURE 7.23 Results of regeneration tests on carbon from a South African plant [50]

However, the choice of screen depends on carbon size, ore particle size, slurry density and viscosity, screen size, and the capital cost for each screening system.

A typical CIP circuit configuration is shown in Figure 7.24. The design and configuration of CIP adsorption systems has been well studied, and several excellent design approaches are available based on empirical models of carbon adsorption processes [18, 53, 541. In the conventional cascade-type system, slurry flows by gravity from one stage


Storage fi7JJ Agitator L

Tank

Carbon

Adsorption Tanks

Slurry from Leach Plant

0

Loaded v 2 - 9 By Hand - Carbon Water I

Elution IT I Anolyte

Steam

Electrowinning Cell

Column

Eluted Carbon

Residue

Gold to Smelting

FIGURE 7.24 Example of a CIP cascade circuit [51]

to the next and carbon is transferred counter-current to slurry flow by pumping or eluc- tion. In the early 1990s, Anglo American Research Laboratories introduced the AARL “pump-cell,” which combines the functions of agitation, screening, and interstage slurry transfer into one compact and highly efficient unit [55]. A unique feature of the pump- cell is that the pump impeller lifts the screened slurry from inside the cylindrical screen and deposits it into the launder to feed the next stage in the circuit. The pumping action generates a pressure differential across the screen surface, promoting slurry flow through the screen. This equipment lends itself to configuration in “carousel” mode, where the counter-current flow of slurry and carbon is accomplished without transfer- ring the carbon from tank to tank, but rather by switching fresh slurry feed from tank to tank. The carbon is removed only for elution and regeneration.

This configuration has the advantages of reducing carbon attrition losses in the adsorption circuit, eliminating short circuiting of carbon during carbon transfer, and potentially improving the stage efficiency and carbon loading profile. Disadvantages include more complex piping design and the need to take a tank off-line to transfer carbon to elution (i.e., less efficient circuit operation during this time) [541. However, a number of large gold plants have utilized the pump-cell CIP technology, including Vaal River No. 2, Hartebeesfontein, Consolidated Murchison, Blyvooruitzicht, West Driefon- tein, East Driefontein, Hoof, and Leeudoorn (all in South Africa), and El Indio (Chile) [54,56]. Of these plants, the largest is Vaal River No. 2 with a throughput rate of approximately 13,000 tpd, indicating that the technology is truly mature.

Either granular or extruded carbons can be used in a variety of size ranges, typically 1.7 x 3.4 mm, 1.2 x 3.4 mm, and 1.2 x 2.4 mm. Carbon concentrations are usually maintained between 5 and 30 g/L, depending on the specific needs of each operation and the carbon properties. High carbon concentrations result in high gold and carbon invento- ries, and increased fine carbon particle losses (and consequently increased gold losses with the fine carbon). The most important factor is the amount of carbon required in the circuit to optimize gold recovery and reduce operating risk, that is, the potential of losing soluble gold to tailings. These requirements vary from operation to operation. Typical


activated carbon consumptions are in the range of 20 to 40 g/t ore, depending on the specific conditions applied and the type and quality of the carbon.

Control of slurry density and viscosity are important in CIP systems, not only for the reasons outlined in Section 7.1.2.1, but also because of the potential for carbon concentration stratification within adsorption tanks, arising from the low apparent density of carbon (0.8 to 0.9 g/mL).

7.1.5.3 Carbon-in-Leach Process CIL is a modification of the CIP process (Section 7.1.5.2) where the leaching and adsorption process steps are performed in the same tanks simultaneously. The process offers advantages of lower capital cost than separate leaching and carbon adsorption systems and can significantly improve gold extraction from ores containing constituents that adsorb gold from leach solutions. In the latter case, the carbon competes with the gold- adsorbing, preg-robbing, or preg-borrowing minerals, and preferentially adsorbs the gold values. However, CIL has some inherent disadvantages compared with CIP, summarized as follows:

1. Larger carbon inventory is required (due to lower loaded carbon concentrations). 2. As a result of item 1, the in-plant gold inventory, or “lockup,” is higher. 3. Fine carbon particles due to carbon attrition and the associated gold losses are

typically higher. 4. Carbon gold loadings are usually lower due to treatment of lower-grade solutions,

which increases the carbon transfer frequency, as well as increasing elution and reactivation requirements.

5. Operating costs are typically higher. Consequently, CIL must be evaluated in detail for each specific application.

AARL pump-cell, carousel-type circuits can be considered in some CIL installations, but generally are only effective if applied in conjunction with a preleach step to provide a higher gold grade to the first-stage CIL (i.e., effectively a partial-leach and CIP circuit rather than true CIL). Circuits that require the use of CIL (due to the presence of preg- robbing constituents in the ore) must generally utilize conventional carbon adsorption slurry tank systems with higher slurry retention times [54].

7.1.5.4 Use of Air or Oxygen in Carbon Adsorption Systems Several carbon adsorption systems, and in particular CIL, sparge air or oxygen into the process tanks to provide dissolved oxygen for gold leaching. (The role of dissolved oxygen in the cyanide leaching reaction is discussed in detail in Chapter 6.) However, air or oxygen can affect carbon adsorption systems in a number of other ways: . Oxygen enhances the adsorption of gold cyanide onto activated carbon (Section

7.1.2.4). . The presence of air and oxygen bubbles in slurry systems can cause the carbon to float, thereby impairing homogeneous mixing. . The oxidation of cyanide by oxygen is catalyzed in the presence of activated carbon (see Chapter 11).

These effects need to be carefully considered for each operation.

7.1.5.5 Carbon-in-Solution Process Activated carbon can be used to extract gold from a wide range of gold cyanide solution streams, including heap or run-of-mine stockpile (dump) leaching solutions, thickener


Loaded Carbon to

i Effluent Solution to Barren Holding Pond

FIGURE 7.25 Flve-stage carbon adsorption circuit for the recovery of gold onto activated carbon from solutlon (CIC)

overflow solutions, unclarified filtrates, and tailings reclaim solutions. Two methods are available for treating solutions: . Fluidized or expanded bed systems . Fixed, packed, or pinned bed systems Several configurations of each have been designed and implemented to meet specific needs, including the following: . Multiple-stage fluidized beds stacked in a single column (e.g., NIMCIX by Mintek) . Multiple-stage fluidized beds as a series of cascading columns . Single, deep fluidized bed column . Up-flowing fixed bed (single or multiple stages) . Down-flowing fixed bed (single or multiple stages) The choice of system depends on the flow rate to be treated, the solution clarity, the gold concentration, and the desired mode of operation. Fluidized beds are preferred for unclarified solutions, where the solids will not rapidly plug the bed. Bed expansions of 10% to 100% are used in practice. Fluidized bed systems have better mass transport properties than packed beds and channeling is usually avoided. A schematic diagram of a multiple-stage, cascading, fluidized bed adsorption system is shown in Figure 7.25.

Packed bed systems are effective for treatment of clean solutions. A gold-loading profile develops across the bed and, because there is no carbon movement, the packed bed columns behave like a plug flow reactor, providing an effective mode of contact. However, channeling of solution through a packed bed is often a major problem in commercial applications, as it causes mass transport rates to be reduced in some regions of the bed, resulting in inefficient adsorption.

7.1.5.6 Elution Several elution systems have been developed and applied on a commercial scale, and these are reviewed here. Typical operating conditions for the different elution systems are summarized in Table 7.4, together with an indication of the applicability of each in industry.

TABLE 7.4 Comparison of typical operating conditions for various carbon elution methods [26, 36, 39, 41, 43, 49, 511

Reagent Scheme Temperature Pressure

Elution type Procedure Presoak Elution ("C) (kPa) 90-1 00 100 None Atmospheric Zadra

(no solvent) 10 g/L NaOH 1-2 g/L NaCN

Pressurized Pressure Zadra

AARL+

Solvent-assisted Zadra/Duval (atmospheric)

Anglo

Murdoch

Micron Research

None

20-50 g/L NaCN 10-20 g/L NaOH

None

20-50 g/L NaCN 10-20 g/L NaOH

80% acetonitrite in H2O

20-50 g/L NaOH 50-1 00 g/L NaCN

10 g/L NaOH 2 g/L NaCN

H2O

1Ooh20% ethanol 10 g/L NaOH 2 g/L NaCN

90% acetone or ethanol in H20

acetonitrite 10 g/L NaCN 2 g/L NaOH

methanol in H-0

20%-40%

60%-80%

135-1 40 400-500

110-120 170-200

80 100

70-90 100

25-70 100

60-80 100

Maximum Gold Time Concentration* (hr) (mglL) Application

36-72 150 Limited use in the United States and Australia. Declining popularity.

8-14

8-14

6-1 2

6-8

8-14

8-80

1,000

1,500

1.500

1,000-2,000

1,500-6,000

3,000-1 0,000

Widely applied in the United States and Australia, to a lesser extent in South Africa. Widely applied in South Africa and throughout the world.The generally preferred elution method.

Very limited application.



Limited application in Australia and South Africa.

* Maximum gold concentrations in solution given for elution of carbon loaded to approximately 4,000-5,000 g/t Au.

t AARL = Anglo American Research Laboratories, South Africa.

0 0 0

330 I THE C

HEM

ISTRY O

F QO

LD EXTRA

CTIO

N

[CH. 7 SEC. 7.11 SOLUTION PURIFICATION AND CONCENTRATION I 331 Atmospheric elution with cyanide and caustic (Zadra process). A solution con-

taining approximately 1% to 2% sodium hydroxide and 0.1% sodium cyanide is used at 95°C and at a flow rate of 2 bv/hr. The process takes between 36 and 72 hr to elute loaded carbon to a low residual loading ( 4 0 0 g/t), equivalent to 100 to 150 bv of solution. Mild steel equipment can be used throughout.

The system is used at elevated temperature (135"C-14OoC) and pressure (400 to 500 kPa) to reduce the elution time to 8 to 14 hr at a flow rate of 2 bv/hr (15 to 30 total bv of solution). Stainless steel elution columns are required.

The carbon is acid washed in dilute mineral acid, water-washed, then soaked in 3% sodium cyanide and 1% to 2% sodium hydroxide solution for approximately 30 min. The carbon is eluted with 6 to 10 bv (at 2 bv/hr) of deionized water at 110°C to 120°C and 70 to 100 kPa pressure [57], Elution is completed in 8 to 14 hr. A butyl rubber-lined elution column is required to withstand acid and alkaline media, which has the drawback of a 113°C maximum operating temperature. In some cases special steel alloys (e.g., Hastelloy) may be preferred to enable operation at higher temperatures.

Several processes use varying proportions of alcohols and glycols to assist in atmospheric elution (i.e., up to 95°C). Systems using 20% of a suitable alcohol (ethanol or methanol) can reduce the elution times of conventional Zadra systems below 12 to 16 hr. Alternatively, glycols (e.g., ethylene or propylene glycol) in proportions of 20% to 25% can reduce elution times to 24 to 36 hr, avoiding the use of alcohols, which are a fire hazard.

Solvent distillation elution. In this process the elution column is configured as a packed bed distillation tower with a solution heater at the base of the column, an over- head condenser, and a reflux pump to recirculate the solvent. The loaded carbon acts as the tower packing. The carbon is presoaked with 1% to 2% sodium hydroxide and 5% to 10% sodium cyanide at ambient temperature. Ethanol, methanol, or acetonitrile is used as the solvent (approximately 0.5 bv). The carbon is refluxed at 65°C to 80°C for 8 hr, with gold values eluted from the carbon by the downflowing condensate. A total of about 1 bv of solution is used to produce a very concentrated solution. The eluted carbon typically has high activity, which is thought to be due to the efficient removal of organic foulants by the hot solvent.

Pressure elution with cyanide and caustic (pressure Zadra process).

Deionized water elution with cyanide presoak (AARL process).

Solvent-assisted elution.

The benefits of using water of low ionic strength for elution have been discussed previously (Section 7.1.3). The presence of dissolved solids not only reduces elution efficiency but also increases the scaling tendency of the solution and may affect its filtration properties. Consequently, water to be used in elution circuits may need to be softened. Plant water containing chlorides should not be used for elution due to the corrosive nature of chlorine gas evolved at the anode during subsequent electrowinning, where applicable [58].

The concentrations of various solution species, which are eluted from carbon more efficiently than they are removed by the subsequent recovery process and operated in parallel with elution, tend to build up. This can have the effect of poisoning the eluate solution, reducing the efficiency of either elution, precious metals recovery, or both. The adverse effect of copper loaded on carbon has been discussed (Section 7.1.3.81, and copper can be removed from carbon prior to gold and silver using a cold elution step. Other examples of eluate poisoning are the following:

Buildup of silica and nickel in the eluate, neither of which are deposited during

. Buildup of zinc ions in the eluate, dissolved during zinc precipitation electrowinning

Next Page